Gemini 2.5 Pro Preview 05 06 Release Details
Google released Gemini 2.5 Pro Preview 05 06 on May 6, 2025. This update improves code and function calling capabilities for the API.
Official Announcement and Scope
Google officially released gemini-2.5-pro-preview-05-06 on May 6, 2025. This date marks the historical event of the model's availability announcement. The source identifies this release as a new version of Google’s most powerful model within the Gemini family. The provider is listed as Google, with input modalities supporting text and image data, while output modality is limited to text.
The documentation explicitly states that the new version includes improvements on code and function calling. It also confirms that gemini-2.5-pro-preview-03-25 will automatically point to the new version of the model. No pricing information, context window size, or maximum output token limits are provided in the source facts.
Model Status and Access Constraints
The source records that open weights are not confirmed for this model. This status indicates that the availability of local installation files or weight downloads is not established by the provided data. It does not confirm whether local deployment is impossible or if API access is the sole method of interaction. The evidence does not describe deployment architecture, integration behavior, or migration requirements.
A deprecation announcement is present in the official changelog, stating that gemini-2.5-pro-preview-05-06 will be shut down on December 2nd. This future date distinguishes the initial release date from the planned end-of-life timeline. The source does not specify which year applies to the shutdown date, nor does it provide details on alternative models for migration.
Practical Evaluation Procedure
Engineers should validate the claimed improvements in code and function calling through structured testing rather than assuming performance gains. A practical evaluation procedure involves creating a baseline dataset of complex coding tasks and multi-step function calling scenarios. Test the model against these inputs to measure accuracy and consistency. Compare the results against the previous version, gemini-2.5-pro-preview-03-25, to isolate specific changes in behavior.
Measure the response quality by checking for syntax errors in generated code and verifying that function calls adhere to defined schemas. Validate that the automatic redirection from the older model ID works as described without requiring manual configuration updates. These steps allow for an objective assessment of the documented features without inferring unverified benefits or architectural advantages.
- Test code generation tasks with varying complexity levels.
- Measure function calling accuracy against strict schema definitions.
- Compare outputs between the 05-06 and 03-25 versions.
- Validate that the old model ID correctly routes to the new version.
Primary sources
Frequently asked questions
Does the source confirm local installation options?
No. The source states that open weights are not confirmed. It does not establish that local installation is impossible, nor does it confirm that API access is mandatory. The available facts do not describe deployment architecture or hosting arrangements.
What specific improvements are documented for this release?
The official changelog documents improvements on code and function calling. It does not provide benchmarks, performance rankings, or detailed explanations of how these features work internally. The source does not claim preserved response speed or other comparative metrics beyond these two areas.